ATIS 0100035-2012 Telepresence Quality of Experience and Quality of Service.pdf

1、 ATIS-0100035 ATIS Standard on - TELEPRESENCE QUALITY OF EXPERIENCE AND QUALITY OF SERVICE As a leading technology and solutions development organization, ATIS brings together the top global ICT companies to advance the industrys most-pressing business priorities. Through ATIS committees and forums,

2、 nearly 200 companies address cloud services, device solutions, M2M communications, cyber security, ehealth, network evolution, quality of service, billing support, operations, and more. These priorities follow a fast-track development lifecyclefrom design and innovation through solutions that inclu

3、de standards, specifications, requirements, business use cases, software toolkits, and interoperability testing. ATIS is accredited by the American National Standards Institute (ANSI). ATIS is the North American Organizational Partner for the 3rd Generation Partnership Project (3GPP), a founding Par

4、tner of oneM2M, a member and major U.S. contributor to the International Telecommunication Union (ITU) Radio and Telecommunications sectors, and a member of the Inter-American Telecommunication Commission (CITEL). For more information, visit .Notice of Disclaimer the lack of noticeable delay; excell

5、ent spatial resolution and motion handling; and enhanced interaction with the remote location. When all this is achieved, the telepresence experience is often described as being “immersive”. The term telepresence can have other, more general, usages that can include things like telerobotics, but the

6、 definition given above clearly limits the meaning here to communications involving people at near life-size images. Also, in this document, the term telepresence is used generically meaning that no specific network, system, or premises equipment platform is assumed. The focus is instead on the appl

7、ication, and the QoE perceived while that application is used. Definitions of other terms used in this document are as follows: Bandwidth: A characteristic of a communication channel that is the amount of information that can be passed through it in a given amount of time, usually expressed in bits

8、per second ATIS Glossary. IP Packet Transfer Delay: The time between the occurrence of two corresponding IP packet reference events, an ingress event (IP packet leaves the source) and an egress event (IP packet arrives at the destination) ITU-T Recommendation Y.1540 see Appendix A. IP Packet Delay V

9、ariation (Jitter): End-to-end 2-point IP packet delay variation (PDV) is defined based on the observations of corresponding IP packet arrivals at the ingress (source) and egress (destination). These observations characterize the variability in the pattern of IP packet arrival events at the egress MP

10、 and the pattern of corresponding events at the ingress MP with respect to a reference delay. The PDV for an IP packet between the sourceand destination is the difference between the absolute IP packet transfer delay of the packet and a defined reference IP packet transfer delay between the source a

11、nd destination ITU-T Recommendation Y.1540 see Appendix A. IP Packet Error Ratio: IP packet error ratio is the ratio of total errored IP packet outcomes to the total of successful IP packet transfer outcomes plus errored IP packet outcomes in a population of interest see Appendix A. IP Packet Loss R

12、atio: IP packet loss ratio is the ratio of total lost IP packet outcomes to total transmitted IP packets in a population of interest ITU-T Recommendation Y.1540 see Appendix A. Motion Handling: For the purposes of this document, video motion is any frame-to-frame change in the spatial image; and mot

13、ion handling is the ability to deliver video motion to users free of commonly occurring temporal artifacts (e.g., “mosquito noise”, “ghosting”, “jerkiness”, and “smearing”). Resolution: A parameter that specifies the ability to distinguish video detail in the spatial dimension or the temporal dimens

14、ion ITU-T Recommendation P.10. Video Frame Jitter: Video Frame Jitter is defined as the difference between the actual arrival time and the expected arrival time of a Video Frame ATIS-0200005. ATIS-0100035 3 4 Acronyms this scenario is included for completeness. Regardless of the configuration of the

15、 transport network(s) used for the IP path, network level (UNI-to-UNI) performance objectives for delay, jitter, and packet loss must be met if the resulting QoE is to be acceptable. These performance objectives have been defined for IP-based networks by ITU-T Recommendation Y.1541 Y.1541 in the for

16、m of six IP QoS classes (see Appendix B)6. Each QoS class consists of network performance objectives (upper bounds) for packet delay, delay variation, and loss, as well as for payload errors within the IP packets. Because only Classes 0 and 1 have an upper bound on packet delay variation, only these

17、 two of the six IP QoS classes in Y.1541 can be meaningfully considered for a demanding application like telepresence. Classes 0 and 1 of Y.1541 differ only in the amount of allowed packet delay (100 and 400 ms, respectively). The allowed delay variation (50 ms), loss (1 103) and errors (1 104) are

18、the same for both classes. While it is desirable to meet the 100 ms network delay objective of Class 0, in practice it is often not possible to do so. Accordingly, the telepresence transport infrastructure should satisfy the network performance objectives of Y.1541 Class 1 (where the allowed network

19、 delay is 400 ms). QoS Requirement R1: The performance of the IP network path should satisfy the ITU-T Recommendation Y.1541 Class 1 upper bounds of 400 ms delay, 50 ms delay variation, 1 103 loss, and 1 104errors. NOTE: Although the Class 1 delay objective has an upper bound of 400 ms, it is strong

20、ly recommended to keep the network delay well below this preferably below 200 ms to assure that there are no noticeable delay effects. With careful transport network planning, delay values below 200 ms are often readily achievable. 6Note that Appendix VIII in ITU-T Recommendation Y.1541 provides use

21、ful information on digital TV. ATIS-0100035 6 It must be emphasized that these Y.1541 objective values are for the entire IP network path (UNI-to-UNI), even if network segments in the path are provided by different operators. In addition, statically allocating these “end-to-end” objectives to indivi

22、dual operators is not feasible because of serious practical and technical obstacles. Different approaches to realizing these objectives across multi-operator paths have been assessed in detail by the ITU-T (see Y.1542, “Framework for achieving end-to-end IP performance objectives”). 8 Telepresence T

23、raffic Characteristics The previous section dealt with the IP network performance that is required to support acceptable telepresence QoE. However, to fully describe what affects telepresence QoE, the traffic characteristics of the telepresence video stream must be understood for two reasons. First,

24、 the realized video quality obviously depends on the spatial resolution and motion rendering each of which ultimately drives the bandwidth required to support the video stream. Second, and perhaps more subtle, is the video frame rate and size generated by the source codec, because it is these video

25、frames that must be serialized as they are transmitted across the underlying IP networks. This section deals with the telepresence traffic characteristics; the next section deals with the video frame aspects of telepresence QoE. Table 1 (see below) illustrates the relationships between spatial resol

26、ution and motion handling aspects, with the associated bandwidth requirements for different telepresence system configurations ATIS-0200005: Table 1 - Telepresence Traffic Characteristics: Max Bandwidth Consumption (kbps) Maximum Bandwidth Consumption Kilobits Per Second (kbps) Resolution 1080p 1080

27、p 1080p 720p 720p 720p 720p Motion Handling Best Better Good Best Better Good Lite Video per Screen (kbps) 4000 3500 3000 2250 1500 1000 936 Audio per Microphone (kbps) 64 64 64 64 64 64 64 Auto Collaborate Video Chan. 500 500 500 500 500 500 100 Auto Collaborate Audio Chan. 64 64 64 64 64 64 64 Sin

28、gle Screen Systems 4628 4128 3628 2878 2128 1628 1164 Total Audio and Video (kbps) 4756 4256 3756 3006 2256 1756 1292 Triple Screen Systems 12756 11256 9756 7606 6256 3766 Total Audio and Video (kbps) + 20% for Layer 2-4 overhead Single Screen Systems max bandwidth (kbps) Tx 5554 4954 4358 3454 2554

29、 1954 1397 Includes layer 2-4 overheard Rx 5707 6107 4607 3607 2707 2107 1550 Triple screen systems max bandwidth (kbps) 15307 13507 11707 9007 6307 4507 Includes layer 2-4 overheard Optional Ad-on Features (kbps) *Not* applicable to 720p Lite 30fps Auto Collaborate 4000 +20% for Layer 2-4 overhead

30、4800 CTRS Recording in CIF 704 +20% for Layer 2-4 overhead 845 SD Interoperability Audio 704 +20% for Layer 2-4 overhead 922 Video 64 WebEx On Touch Audio 304 +20% for Layer 2-4 overhead 442 Video 64 ATIS-0100035 7 This table demonstrates bandwidth usage for two levels of resolution (1080p and 720p)

31、 each with associated levels of motion handling (“Best”, Better”, and “Good”). The bandwidth requirement to avoid potential QoE impairments during the telepresence session is as follows: QoE Requirement R1: The bandwidth required across the end-to-end path should take into account all necessary fact

32、ors for enabling the desired levels of telepresence service for resolution and motion handling. NOTE: This requirement has two implications: First, if it is known that a limited amount of bandwidth is available (for example, due to the access network), then the resolution and motion handling combina

33、tion chosen must not exceed the limited bandwidth. Conversely, if it is known that bandwidth is not constrained by the end-to-end transport path, then the desired combination of resolution and motion handling can be made independently from bandwidth considerations as long as that bandwidth is guaran

34、teed. Of course cost may then be a factor. 9 Video Frame Jitter Considerations Now that telepresence traffic characteristics have been described, the QoE-affecting aspect of video frame jitter can be meaningfully examined. Unlike IP network impairments, the effects of not adequately handling video f

35、rame jitter directly impacts telepresence QoE. This is because of the direct influence on the video decoder, and thus the associated visible image impairments (gradual screen repair, full screen refreshes, picture freeze, etc.). The first important consideration in understanding video frame jitter i

36、s knowing that the video frame rate is 30 frames/s, so that a video frame is generated by the source codec every 33 ms. However, these video frames can vary greatly in size (from 1 to 65 Kbytes, with the average being 16 Kbytes), depending on the image complexity (e.g., spatial and temporal redundan

37、cy). These varying-sized video frames are packetized and then transmitted across facilities operating at widely varying underlying clock rates (e.g., T1, 10 Mb/s Ethernet, T3, etc.). These clock rates, combined with the video frame size, determine how much time it takes to completely transmit each f

38、rame serialization delay-and clearly this time will vary as the frame size varies. Thus video frame jitter is generated, as the following figure graphically depicts ATIS-0200005: ATIS-0100035 8 Figure 2 - Understanding Video Frame Jitter versus IP Packet Jitter The previous figure shows an example o

39、f how video frame jitter is generated. It also shows that the frame buffer in the receive telepresence system must have enough storage to readily accommodate this jitter, as the frames will not arrive when “expected” yet every 33 ms it has to provide the next frame (independent of size) to the decod

40、er. To appreciate what telepresence systems have to do to accommodate possible ranges of video frame jitter, the following table shows maximum jitter as a function of clock rate and frame size ATIS-0200005: ATIS-0100035 9 Table 2 - Understanding Video Frame Jitter Versus IP Packet Jitter Per Screen

41、T1 E1 4 x T1 4 x E1 10Mbps Ethernet E3 T3 OC3 OC12 Circuit Bit Rate (kbps) 1, 544 2, 048 6, 176 8,192 10,000 34,368 44,736 155,520 622,080 Max Frame Size 65,000 Bytes Average Frame Size 16,000 Bytes Min Frame Size 1,000 Bytes Max-Frame Serialization Rate (MFSR) 337 ms 265 ms 84 ms 64 ms 52 ms 15 ms

42、12 ms 4 ms t1) and (t2 t1) Tmax. If the packet is fragmented within the NSE, t2is the time of the final corresponding egress event. The end-to-end IP packet transfer delay is the one-way delay between the MP at the SRC and DST as illustrated in Figure A.1. Figure A.1 - IP packet transfer delay event

43、s A.2 End-to-end 2-point IP Packet Delay Variation (IPDV) The variations in IP packet transfer delay are also important. Streaming applications might use information about the total range of IP delay variation to avoid buffer underflow and overflow. Extreme variations in IP delay will cause TCP retr

44、ansmission timer thresholds to grow and may also cause packet retransmissions to be delayed or cause packets to be retransmitted unnecessarily. End-to-end 2-point IP packet delay variation (PDV) is defined based on the observations of corresponding IP packet arrivals at ingress and egress MP (e.g.,

45、MPDST, MPSRC). These observations characterize the variability in the pattern of IP packet arrival events at the egress MP and the pattern of corresponding events at the ingress MP with respect to a reference delay. The 2-point PDV (vk) for an IP packet k between SRCand DST is the difference between

46、 the absolute IP packet transfer delay (xk) of packet k and a defined reference IP packet transfer delay, d1,2, between those same MPs (see Figure A.2): vk= xk d1,2. ATIS-0100035 12 Figure A.2 - 2-point IP packet delay variation The reference IP packet transfer delay, d1,2, between SRC and DST is th

47、e absolute IP packet transfer delay experienced by a selected IP packet between those two MPs. Positive values of 2-point IPDV correspond to IP packet transfer delays greater than those experienced by the reference IP packet; negative values of 2-point PDV correspond to IP packet transfer delays les

48、s than those experienced by the reference IP packet. The distribution of 2-point PDVs is identical to the distribution of absolute IP packet transfer delays displaced by a constant value equal to d1,2. A.3 Using Minimum Delay as the Basis for Delay Variation As illustrated in Figure A.2, the delay v

49、ariation of an individual packet is naturally defined as the difference between the actual delay experienced by that packet and a nominal or reference delay. The preferred reference (used in ITU-T Y.1541 IPDV objectives) is the minimum delay of the population of interest. This ensures that all variations will be reported as positive values, and simplifies reporting the range of variation (the maximum value of variation is equal to the range). Distributions of delay variation in IP networks often exhibit a bias toward the min